Re: corrupt data - caused by lm-sensors?

On 14 Jul 2003 14:27:33 -0700
Will LaShell <will(at)lashell(dot)net> wrote:

> On Sat, 2003-07-12 at 12:43, Zeno Davatz wrote:
> > Robert Treat writes:
> >
> > > On Fri, 2003-07-11 at 01:58, Zeno R.R. Davatz wrote:
> > >> Hi List
> > >>
> > >> Out of some reason our data of our postgresql database has been corrupted. When we try to connect to the database we get:
> > >>
> > >> psql: ERROR: _mdfd_getrelnfd: cannot open relation pg_type_oid_index:
> > >> No such file or directory
> > >>
> > >> REINDEX does not work. We get the same error.
> > >>
> > >> pgfsck gives us: wrong blockseize.
> > >>
> > >> We are using postgresql 7.3.3-1 (debian).
> > >>
> > >
> > > uh oh, sounds like you might have some flakey hardware.
> > >
> > > try vacuumdb template1 and if that gets you in you might want to do a
> > > pg_dump and then do some hardware testing.
> > Thanks for the hint. In the meantime we could get access to some data with
> > pgfsck 0.14 - great tool, great maintainer, very dedicated (Tom Lane says
> > pgfsck does not exist).
> >
> > What I forgot to mention:
> > I installed lm-sensors (cvs) version and force-loaded some drivers. Then
> > sensors-detect gave me a segmentation fault. Do you think that could have
> > caused the data-corruption?
>
> You did this on a production system? At any rate, depending on what /
> how it segfaulted it has a good chance of confirming what the others
> have suggested in that you have some bad hardware on that machine. Note
> that this shouldn't have caused the drive corruption unless the segfault
> took out a piece of the disk io kernel subsystem. Although I have to say
> that using development versions of software on production machines is
> not necessarily the most safe practice one can engage in.

Thanks for the info. Do you know of any tool to check my hardware especially the hdd's?

Thanks for feedback.

Zeno